> For the complete documentation index, see [llms.txt](https://carec.gitbook.io/carec-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://carec.gitbook.io/carec-docs/documentation/simulation/synthetic-data-and-ai-evaluation.md). # Synthetic Data and AI Evaluation Controlled data generation and validation for CAREC perception and AI features. Simulation can create labeled scenes for perception development, but synthetic performance does not establish real-world safety. ## Generation dimensions * Lighting, exposure, glare, shadows, and time of day * Camera pose, lens model, depth noise, motion blur, and occlusion * Room, corridor, doorway, ramp, surface, and clutter variation * People, mobility aids, pets, furniture, and unexpected obstacles * Positive, negative, rare, and ambiguous safety events ## Dataset manifest Every generated dataset needs simulator and world versions, asset licenses, parameter ranges, seeds, class definitions, annotation schema, split method, known gaps, and generation commit. ## Leakage and balance controls Split by world and asset family—not only by frame—so near-duplicate scenes do not inflate results. Report class balance, scenario coverage, failure slices, confidence calibration, and false-clear events. ## Reality-gap evaluation Validate models on separately governed real data, compare per-condition performance, document synthetic-to-real gaps, and keep conservative fallbacks. AI output should feed the safety architecture through defined interfaces; it must not silently become the sole authority for motion. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://carec.gitbook.io/carec-docs/documentation/simulation/synthetic-data-and-ai-evaluation.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.